Watch the recording and ask questions about this event in this thread!
- When: 15 November 2022 at 17:00 (UTC)
- Speaker(s): Vladislav Vávra
- Talk(s): Australian Virtual Power Plant running on Kafka Streams
Energy production is moving towards distributed energy sources these days. One of the sources are household photovoltaic power plants (PV). Besides obvious economical and ecological benefits they also represent challenges to the utility grid stability. One way to overcome such challenges is to install batteries at homes and dispatch the energy from them to keep the grid stable. At AmpX we are working on a platform for an Australian customer for monitoring and managing hundreds of household PVs. Those PVs are aggregated in a virtual power plant (VPP) in order to stabilize the grid and maximize the households revenues. As the data generated by the PVs is a never-ending stream of many metrics we opted to use Kafka Streams for the processing. The story starts by polling the data from the provider of the PV systems to the Kafka in the Avro format. Then we perform various aggregations like 5 minutes bucketing or day bucketing among multiple time zones. Based on the data we take various actions. The data is also stored in a Postgres database using Kafka Connect. A web server queries Postgres and allows the users to analyze the state of the VPP. This approach allows us to split the data pre-processing, control and the user queries and this increases system stability and makes the developers live easier.